LSM configuration lost

Didier Godefroy · *nix forums beginner Joined: 28 Feb 2005 Posts: 7

Hello all,

I'm in a sticky situation with my LSM configuration on an Alpha800 with 4
drives and tru64 5.1
The drives dsk0 and dsk1 are a mirror and have everything on them, system
and user files, there's only one partition used on dsk2 and it's nothing
useful, then dsk3 was supposed to be a spare.
I recently had a drive failure on dsk1 (dsk3 was not there at the time to be
used as a spare), so the system kept on going on dsk0 until I replaced dsk1
and manually re-created the mirroring, then added dsk3 as a spare. However I
wasn't able to create extra LSM configuration copies and the only one was on
dsk0 and some glitch must have happened with dsk0 which caused a reboot.
The system tries to boot normally from dsk0 and then can't find osf_boot and
stops. I manually force a boot from dsk1 and it only boots to single user
and fails bcheckrc during the disk naming and LSM can't start because no
configuration copy can be found.
I can't mount and other filesystems, so I don't have access to any useful
utility to help fix this and no access to a saved LSM config copy. All I
have is the root filesystem available.
All drives are 18gigs and basically only dsk0 and dsk1 are used, what is on
dsk2 isn't vital and dsk3 is just a hot spare.
I tried a few things but none worked because I don't have access to most
commands.
What could I try to start restoring this???

--
Didier Godefroy
mailto:dg@ulysium.net

Didier Godefroy · *nix forums beginner Joined: 28 Feb 2005 Posts: 7

The recovery process was rather long but it turned out fine.
I couldn't fix anything with the system booted as it was in single user with
only the root filesystem mountable.
I ended up installing a separate system on an other drive. From there, I
could mount the root partition and fix things to make it bootable again.

I disabled LSM on all the volumes manually by changing the links in
/etc/fdmns, pointing them back to their original advfs devices. I edited
sysconfigtab and inittab to take out LSM there as well.

Once LSM was disabled, I was able to boot that system again, although not
without some errors. From that booted state, it was possible to further fix
the small problems causing errors.

I had removed the dsk0 from the mirror, only keeping the dsk1, which I
booted from.

Once most errors were fixed, I re-enabled LSM and re-encapsulated that dsk1
and rebooted. It was just a matter of re-building the mirror to get it all
back to a state like it was before all that trouble.

There was only one error that remained and couldn't fix, although I think it
fixed itself once that system was connected again to the network. I was the
evm daemon that was complaining about not being able to communicate with
syslog and binlog. Both syslog and binlog were working but evm just couldn't
talk to them, but I guess it came back to normal once the network was on.

Google